backend-debug

Pass

Audited by Gen Agent Trust Hub on Nov 9, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data which could potentially contain malicious instructions intended to influence the agent's behavior during debugging tasks.
  • Ingestion points: User-provided bug reports, error logs, stack traces, and data samples (SKILL.md, Step 1; triage-playbook.md).
  • Boundary markers: The skill uses natural language instructions to separate steps but lacks explicit technical delimiters for all ingested data.
  • Capability inventory: The skill has capabilities to modify the filesystem (Apply fix, Step 5), execute shell commands for testing and reproduction (Step 2, Step 3), and manage git repositories (Step 4).
  • Sanitization: There are explicit instructions to redact secrets and PII from logs and reports (Step 2, Step 4, Step 7), though this is focused on data privacy rather than preventing instruction injection.
Audit Metadata
Risk Level
SAFE
Analyzed
Nov 9, 2026, 09:36 AM